Search Results

To prevent the severe consequences of unsafe driving behaviors, it is crucial to monitor and analyze the state of the driver. Developing an effective driver state monitoring (DSM) systems is particularly challenging due to limited computation capabilities of embedded systems in automobiles and the need for finishing processing in real-time. However, most of the existing research work was conducted in a lab environment with expensive equipment while lacking in-car benchmarking and validation. In this paper, a DSM system that estimates driver's alertness and drowsiness level as well as performs emotion detection built with a cost-effective embedded system is presented. The proposed system consists of a mono camera that captures driver's facial image in real-time and a machine learning based detection algorithm that detects facial landmark points and use that information to infer driver's state.

In this paper, a newly developed three-dimensional (3D) bird's-eye view map drawing technique for car navigation systems is described. Conventional navigation systems give pseudo-perspective views which can not express ruggedness like hills and valleys. Our newly developed navigation system can display undulation of the land from viewpoints above and behind the current position, so that ups and downs of roads along with the driver's destination can be seen easily. The 3D-road map is not only effective during navigation but also during route planning, because it assists in searching for fine views before travel. In order to achieve the 3D-map view, we developed graphics software libraries, which work on a 32-bit RISC processor and on a low-cost graphics accelerator LSI with texture mapping capability. The graphics software libraries are constructed with three stages, the perspective projection stage, visible-surface determination stage, and rendering stage.

The problem of high fatal accident rates due to drunk driving persists, and must be reduced. This paper reports on a prototype system mounted on a car mock-up and a prototype portable system that enables the checking of the drivers’ sobriety using a breath-alcohol sensor. The sensor unit consists of a water-vapor-sensor and three semiconductor gas sensors for ethanol, acetaldehyde, and hydrogen. One of the systems’ features is that they can detect water vapor from human-exhaled breath to prevent false detection with fake gases. Each gas concentration was calculated by applying an algorithm based on a differential evolution method. To quickly detect the water vapor in exhaled breath, we applied an AC voltage between the two electrodes of the breath-water-vapor sensor and used our alcohol-detection algorithm. The ethanol level was automatically calculated from the three gas sensors as soon as the water vapor was detected.

We have developed a model-based control for the air intake system in a variable valve engine, employing total engine simulation, the response surface method and multi-objective optimization scheme. In our technique, we performed the simulation model tuning and validation, followed by the creation of a dataset for the polynomial regression analysis of the charging efficiency. A D-optimal design, robust least squares method, and likelihood-ratio test were demonstrated to yield a robust and accurate control model. Coupling the total engine simulator with a genetic algorithm, model based calibration for optimal valve timing stored in lookup table was carried out under multiple objectives and restrictions. The reliability of the implementation control model, which considers the effect of gas dynamics in the intake system, was confirmed using a model-in-the-loop simulation.

The purpose of this study is to develop model-based methodologies which employ thermo-fluid dynamic engine simulation and multiple-objective optimization schemes for engine control and calibration, and to validate the reliability of the method using a dynamometer test. In our technique, creating a total engine system model begins by first entirely capturing the characteristics of the components affecting the engine system's behavior, then using experimental data to strictly adjust the tuning parameters in physical models. Engine outputs over the full range of engine operation conditions as determined by design of experiment (DOE) are simulated, followed by fitting the provided dataset using a nonlinear response surface model (RSM) to express the causal relationship among engine operational parameters, environmental factors and engine output. The RSM is applied to an L-jetronic® air-intake system control logic for a turbocharged engine.

In recent years, integrated vehicle control systems have been developed to improve fuel economy and safety. As a result, engine control is shifting to torque-based systems for throttle / fuel / ignition control, to realize an engine torque demand from the system. This paper describes torque-based engine control technologies for SI (Spark Ignition) engine to improve torque control accuracy using a feedback control algorithm and an airflow sensor.

We have developed a full virtual engine system prototyping platform with 4-cylinder engine plant model, SH-2A CPU hardware model, and object code level software including OSEK OS. The virtual engine system prototyping platform can run simulation of an engine control system and digital knock detection system including 64-pt FFT computations that provide required high-resolution DSP capability for detection and control. To help the system design, debugging, and evaluation, the virtual system prototyping consists of behavior analyzer which can provide the visualization of useful CPU internal information for control algorithm tuning, RTOS optimization, and CPU architecture development. Thus the co-simulation enables time and cost saving at validation stage as validation can be performed at the design stage before production of actual components.